Restart File Questions

[dhafner]

HI,

I am working on my first unsteady flow model and am having major instability problems. I'm trying to model a huge retention basin that collects bank overflows from a nearby the creek.

I'd like to to create a restart file for it, but am not sure if I'm heading in the right direction.

-I've simplified the model, removing all bridges, weirs, and culverts.
-I'm attempting to stabilize the model with an u/s constant flow hydrograph and a d/s high tailwater stage hydrograph that ramps down to a normal tailwater elevation.

I can stabilize the model, but the constant flow is 2500 cfs (the basin will normally be empty) and the computation interval is 30 minutes. When ever I lower the time interval, the model becomes unstable.

What constant flowrate should I probably use in this case where there normally is no flow?

How do I determine if my computation time interval is too high?

Should I stabilize the constant flow model first (minus the structures) before adding the structures back into it? Then attempt to stabilize it again?

Thanks for your patience in reading this. I'm very frustrated and looking for any help to solve these problems.

Thank you.

Dave

[jarvus lunt]

I've experimented with the approach of making a restart file by starting with a high tailwater and gradually lowering it down. I've not had a lot of luck with that approach, but I know that some users have. I think you need to lower things really slowly.

After you have created a restart file, you can't change the geometry. So if you are trying to create a restart file, leave all the structures in place.

However, you may not really need a restart file. Stability problems can be tricky, especially for new users. If you fix underlying problems, you may find that you don't need a restart file at all.

When trying to build a stable model, I generally leave the structures in. If I was going to remove structures, I would probably remove lateral structures before bridges/culverts. For a large data set, I will sometimes remove a large chunk and work on a smaller section--for instance, the downstream most section.

Unless the channels are huge, 2500 cfs for an area that is normally dry sounds rather high. You might want to investigate "pilot channels" to see if you can use a smaller base flow. 30 min time step also sounds on the large side, unless it is a fairly large, sluggish system. If reducing the time step significantly changes the results (or causes its own stability issues) then the time step is probably too large.